File size: 1,401 Bytes
c30a3ac |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 |
import json
import requests
from typing import List
from tqdm import tqdm
from langchain.embeddings.base import Embeddings # adjust if using another base class
class CustomAPIEmbeddings(Embeddings):
def __init__(self, api_url: str, show_progress: bool = True, batch_size: int = 32):
self.api_url = api_url
self.show_progress = show_progress
self.batch_size = batch_size
def embed_documents(self, texts: List[str]) -> List[List[float]]:
lst_embedding = []
iterator = range(0, len(texts), self.batch_size)
iterator = tqdm(iterator) if self.show_progress else iterator
for i in iterator:
batch = texts[i: i + self.batch_size]
payload = json.dumps({"inputs": batch})
headers = {'Content-Type': 'application/json'}
try:
response = requests.post(self.api_url, headers=headers, data=payload)
embeddings = json.loads(response.text)
lst_embedding.extend(embeddings) # assumes response is a list of embeddings
except Exception as e:
print(f"Error on batch {i // self.batch_size}: {e}")
print(response.text if response else "No response")
return lst_embedding
def embed_query(self, text: str) -> List[float]:
return self.embed_documents([text])[0] |